Description of problem: If grub is installed onto the '/boot' partition on a dual boot system, its use of the "makeactive" option on the other OS renders grub disabled if the user ever starts the other OS. The grub info page says that this option isn't all that necessary anymore. How reproducible: the one time I've tried it Steps to Reproduce: Install onto a machine with an existing non-Linux OS. /dev/hda1: Win2K (anything non-Linux will probably work) /dev/hda2: swap /dev/hda3: /boot /dev/hda5: / Generated file: # NOTICE: You have a /boot partition. This means that # all kernel paths are relative to /boot/ default=0 timeout=30 splashimage=(hd0,2)/grub/splash.xpm.gz title Red Hat Linux (2.4.6-2) root (hd0,2) kernel /vmlinuz-2.4.6-2 ro root=/dev/hda5 hdd=none initrd /initrd-2.4.6-2.img title Windows rootnoverify (hd0,0) makeactive chainloader +1
Chang
Where exactly in the info pages does it say not to use makeactive? I'm not seeing it with a quick double check of looking although I see multiple places where it says to boot DOS/Windows that you should use makeactive. I know we've done some basic testing with a DOS partitions that have worked with this.
Start with 'info grub', go to "General Boot Methods", "Chain-loading", and then check the footnote referenced in step 2. It says "This is not necessary for most of the modern operating systems." Frankly, I never had to do this with LILO to boot any previous versions of Windows that I was forced to live with. I just verified this on another machine with Windows 2000 on /dev/hda1 and Red Hat 7.1 on /dev/hda2 . When you did the tests with DOS partitions, was GRUB installed to the MBR? I would expect it to work correctly, then.
Hrmm... it's still a bit unclear from the documentation to me, but I'm mostly convinced by the discussion on testers-list and seeing that it still works in the mbr without makeactive that it's at least not more incorrect. Changed to not use makeactive in cvs